the relationship between epicardial fat and indices of obesity
DESCRIPTION
Descripción sobre el papel de la grasa epicardica y los factores de riesgos cardiovasculares implicados en la obesidad y como este indice de grasa epicardica establece relación directa con mortalidad de causa cardiaca en obesos sean diabéticos o noTRANSCRIPT
The Relationship Between Epicardial Fatand Indices of Obesity and the Metabolic Syndrome:
A Systematic Review and Meta-Analysis
Simon W. Rabkin, MD, FRCPC, FACC
Abstract
Epicardial fat (epicardial adipose tissue, EAT) has been implicated in the pathogenesis of coronary artery disease(CAD). The objective of this study was to examine the relationship between EAT and generalized obesity, centralor visceral adipose tissue (VAT), and the components of the metabolic syndrome—systolic blood pressure (SBP),triglycerides (TGs), high-density lipoprotein cholesterol (HDL-C), and fasting blood glucose (FBG)—that arelinked to CAD. A systematic review of the literature, following meta-analysis guidelines, was conducted untilMay, 2013, using the search strategy ‘‘Obesity’’ OR ‘‘abdominal obesity’’ OR ‘‘metabolic syndrome’’ OR ‘‘met-abolic syndrome X’’ AND ‘‘epicardial fat’’. Thirty-eight studies fulfilled the criteria. There was a highly signif-icant (P < 0.00001) correlation between EAT and body mass index (BMI), waist circumference (WC), or VAT. Thecorrelation between EAT and VAT was significantly (P < 0.0001) greater than the correlation between EAT andWC, which in turn was significantly greater than the correlation between EAT and BMI. Overall, EAT was7.5 – 0.1 mm in thickness in the metabolic syndrome (n = 427) compared to 4.0 – 0.1 mm in controls (n = 301). EATcorrelated significantly (P < 0.0001) with SBP, TGs, HDL, and FBG, but the strength of the association was lessthan one-half of the relationship of EAT to indices of obesity. The results of multivariate analysis were lessconsistent but show a relationship between EAT and metabolic syndrome independent of BMI. In summary, thevery strong correlation between EAT and VAT suggests a relationship between these two adipose tissue depots.Measurement of EAT can be useful to indicate VAT. Whereas EAT correlates significantly with each of thecomponents of the metabolic syndrome— SBP, TGs, HDL, or FBG—the magnitude of the relationship is con-siderably and significantly less than the relationship of EAT to BMI. These data show the strong relationshipbetween EAT and BMI but especially with WC and VAT. They also demonstrate the smaller magnitude of theassociation of EAT with standard coronary risk factors, related to the metabolic syndrome, and suggest that theunique features of this adipose tissue warrant detailed investigation.
Introduction
Epidemiologic data linking obesity to ischemic heartdisease has been recognized for a long time, but the
underlying mechanism of this association had not beenclearly defined.1 Epicardial fat (epicardial adipose tissue,EAT) is the adipose tissue depot located mainly around theepicardial coronary vessels and is also present on the myo-cardial surface, from which it can extend into the heart to beinterspersed with myocardial muscle fibers.2 Recent evi-dence suggests that while EAT serves several importantphysiologic functions,2 excess EAT is associated with coro-nary artery atherosclerosis.3 The question arises whether the
relationship of EAT to coronary artery disease (CAD) eventscan be explained by the association of EAT with generalizedobesity, abdominal obesity, and other obesity-related factorsthat have been linked to cardiovascular disease (CVD),namely blood pressure, fasting glucose, and dyslipidemia.4–7
The constellation of factors that define the metabolic syn-drome—abdominal obesity, and alterations of blood pres-sure, fasting glucose, and serum lipids8,9—are associatedwith an increased risk for the development of CVD.10,11
Considering that EAT has been linked to CAD, the objectiveof this study is an in-depth examination of the relationship ofEAT to obesity and specifically to central abdominal obesityor visceral adipose tissue (VAT) as well as to the individual
Department of Medicine (Cardiology), University of British Columbia, Vancouver, British Columbia, Canada.
METABOLIC SYNDROME AND RELATED DISORDERSVolume 12, Number 1, 2014� Mary Ann Liebert, Inc.Pp. 31–42DOI: 10.1089/met.2013.0107
313131
components of the metabolic syndrome, which are tradi-tional risk factors for CAD, and their combination as themetabolic syndrome.
Methods
Search strategy
A systematic review of the literature was conducted fol-lowing meta-analysis guidelines and using a preferred re-porting system for systematic review and meta-analysis forobservational studies.12,13 A systematic search was con-ducted until May, 2013, to identify studies that examined therelationship between obesity and EAT. The Medline andEMbase databases were searched using the PubMed andOvidSP platforms. The full electronic search strategy usedwas ‘‘Obesity’’ OR ‘‘abdominal obesity’’ OR ‘‘metabolicsyndrome’’ OR ‘‘metabolic syndrome X’’ AND ‘‘epicardialfat’’ limited to humans. Similar search terms were used whensearching OvidSP Medline and OvidSP EMbase. The Co-chrane was queried for reviews on EAT.
Eligibility criteria
Studies that met the following criteria were included: (1)An original study published in a peer-review journal, (2)measurement of EAT in persons with or without obesity orthe metabolic syndrome, (3) adults (aged ‡ 18 years), (4)subjects without a secondary cause of metabolic syndromesuch as human immunodeficiency virus (HIV), (4) presen-tation of a univariate correlation between EAT and a mea-sure of body weight or obesity, i.e., body mass index (BMI) orwaist circumference (WC) or visceral adiposity/abdominalfat. Duplicate studies, non-English studies, abstracts fromunpublished studies, reviews, case reports, and letters wereexcluded. In a few cases, studies from the same author in thesame time frame were included because it was not possibleto confirm that they were duplicate studies of the same pa-tient population. Studies of paracardiac fat were excludedfrom review both because these studies are not on EAT andthe potential confusion that might result combining parietalpericardial fat with EAT.
Data extraction
A predefined protocol was used in accordance with rec-ommendations.12 From each eligible study, patient charac-teristics and the method of measurement of EAT wererecorded systematically. Patient characteristics included av-erage age, sex, correlation with BMI, WC, abdominal VAT,blood pressure, fasting blood glucose (FBG), triglycerides(TGs), and high-density lipoprotein cholesterol (HDL-C). Thecriteria for the diagnosis of the metabolic syndrome werealso recorded.
Outcome assessment
The principal summary outcome measure was the uni-variate correlation between EAT and BMI, WC, and thecomponents of metabolic syndrome, namely FBG, systolicblood pressure (SBP), TGs, and HDL-C. For each eligiblefinding, the correlation coefficient, direction of correlation,and significance were recorded. The other outcome mea-surement was also the measurement of EAT in persons withor without the metabolic syndrome and multivariate analysis
relating EAT to metabolic syndrome considering otherfactors.
Statistical analysis
Meta-analyses of the aggregate patient data were con-ducted with the Comprehensive Meta-analysis Version 2(Biostat, Englewood, NJ). From each entry, the study name,sample size, mean and standard deviation, and correlationcoefficient were entered. The Fischer z transformation wasused to compare the summary correlation coefficients of thedifferent techniques.14 To assess heterogeneity, the CochraneQ statistic was calculated. Statistical significance was set asP < 0.05.
To determine if any one publication had a disproportionateeffect on the summary correlation coefficient, sensitivity ana-lyses were conducted (Comprehensive Meta-analysis Version2). Each study was sequentially removed and the analysis wasrepeated. The point estimates of correlation coefficients werecompared to the primary result to identify any changes insignificance of the correlation.
Results
There were 38 studies that fulfilled the eligibility criteria(Fig. 1). The studies represented were predominantly fromcountries in Europe, but other countries throughout theworld were included (Table 1). There were 26 studies thatpresented data with the univariate correlation coefficientbetween EAT and BMI (Fig. 2). Twenty-three studies re-ported a significant correlation; two studies reported nosignificant relationship and one was almost significant at the5% level. Overall there was a highly significant (P < 0.00001)correlation between EAT and BMI. The total sample size was2490 persons.
Sensitivity analysis showed that exclusion of any singlestudy from the analysis did not significantly alter the overallfindings. In addition, there were four studies that presentedthe data as a significance level but without correlation coef-ficients. Four studies reported a significant correlation be-tween EAT and BMI,15–17,53 and one study did not find asignificant relationship.18 Thus, 87% (26 of 30) of studiesfound a significant relationship between EAT and BMI.
FIG. 1. Flow diagram for selection of studies that wereincluded in the meta-analyses.
32 RABKIN
Ta
bl
e1.
St
ud
ie
sE
va
lu
at
ed
,C
ou
nt
ry
of
Au
th
or
s,
Sa
mp
le
Siz
e,
Po
pu
la
tio
nA
ge
,S
ex
,
an
dM
et
ho
do
fM
ea
su
re
me
nt
of
Ep
ic
ar
dia
lF
at
Tissu
ea
nd
It
sA
sse
ssm
en
t
Au
thor
Cou
ntr
yN
Ag
e(y
r)M
en(%
)D
efin
itio
nm
etab
olic
syn
dro
me
Met
hod
epic
ard
ial
fat
Ex
pre
ssio
n
Gra
ner
etal
.20
131
9F
inla
nd
7743
100
Har
mo
niz
edD
efin
itio
nM
RI
Are
aY
org
un
etal
.20
134
4T
urk
ey83
5643
Ad
ult
Tre
atm
ent
Pan
elII
IC
TT
hic
kn
ess
To
ket
al.
2013
48
Tu
rkey
9047
54A
du
ltT
reat
men
tP
anel
III
Ech
oT
hic
kn
ess
Bo
rru
elet
al.
2013
20
Sp
ain
106
2924
No
ne
Ech
oT
hic
kn
ess
Ker
ret
al.
2013
52
Can
ada
9464
56A
du
ltT
reat
men
tP
anel
III
CT
Vo
lum
eF
uet
al.
2013
,38
Lia
ng
etal
.20
124
9T
aiw
an32
4410
0A
du
ltT
reat
men
tP
anel
III
MR
IT
hic
kn
ess
Nak
azat
oet
al.
2012
37
US
A37
458
59N
on
eC
TV
olu
me
Kil
icas
lan
etal
.20
122
1Ja
pan
9631
0N
on
eE
cho
Th
ick
nes
sM
azzo
cco
liet
al.
2012
62
Ital
y87
7452
No
ne
Ech
oT
hic
kn
ess
Sh
emir
ani
and
Kh
osh
avi
2012
63
Iran
292
5660
No
ne
Ech
oT
hic
kn
ess
Cav
alca
nte
etal
.20
122
2U
SA
110
5565
Ad
ult
Tre
atm
ent
Pan
elII
IE
cho
Th
ick
nes
sG
abo
rit
etal
.20
121
8F
ran
ce23
4121
No
ne
MR
IV
olu
me
Sir
on
iet
al.
2012
23
Ital
y11
352
76N
AM
RI
Are
aG
ran
eret
al.
2012
24
Fin
lan
d48
28N
AN
on
eE
cho
Th
ick
nes
sG
abo
rit
etal
.20
122
5F
ran
ce66
3536
No
ne
MR
IV
olu
me
Yer
ram
asu
etal
.20
122
6U
K33
354
62H
arm
on
ized
Defi
nit
ion
CT
Vo
lum
eL
aiet
al.
2012
17
Tai
wan
359
5269
Ad
ult
Tre
atm
ent
Pan
elII
IE
cho
Th
ick
nes
sB
ette
nco
urt
etal
.20
121
5U
K21
558
61N
on
eC
TV
olu
me
Kar
adag
etal
.20
112
7T
urk
ey12
056
33In
tern
atio
nal
Dia
bet
esF
eder
atio
nE
cho
Th
ick
nes
sS
eng
ul
and
Du
man
2011
45
Tu
rkey
7649
55A
du
ltT
reat
men
tP
anel
III
Ech
oT
hic
kn
ess
Pie
rdo
men
ico
etal
.20
114
6It
aly
174
6142
Ad
ult
Tre
atm
ent
Pan
elII
IE
cho
Th
ick
nes
sIa
cob
elli
set
al20
112
8C
anad
a22
6110
0N
on
eE
cho
Th
ick
nes
sS
icar
iet
al.
2011
29
Ital
y49
4878
No
ne
MR
IT
hic
kn
ess
Mal
avaz
os
etal
.20
104
2It
aly
2042
100
No
ne
Ech
oT
hic
kn
ess
Ay
din
etal
.20
103
1T
urk
ey50
52N
AA
du
ltT
reat
men
tP
anel
III
Ech
oT
hic
kn
ess
Kim
etal
.20
101
6Ja
pan
101
4910
0N
AE
cho
Th
ick
nes
sS
tram
agli
aet
al.
2010
41
Ital
y55
7453
Ad
ult
Tre
atm
ent
Pan
elII
IE
cho
Th
ick
nes
sW
ang
CP
etal
.20
095
0T
aiw
an12
755
80A
du
ltT
reat
men
tP
anel
III
CT
Vo
lum
eW
ang
TD
etal
.20
093
2T
aiw
an14
759
85A
du
ltT
reat
men
tP
anel
III
CT
Th
ick
nes
s,ar
ea,
vo
lum
eG
ort
eret
al.
2008
64
Net
her
lan
ds
6061
73N
on
eC
TT
hic
kn
ess,
vo
lum
eO
ky
ayet
al.
2008
51
Tu
rkey
246
5438
Inte
rnat
ion
alD
iab
etes
Fed
erat
ion
sE
cho
Th
ick
nes
sIa
cob
elli
set
al.
2008
a34
Can
ada
57a
4870
Ad
ult
Tre
atm
ent
Pan
elII
IE
cho
Th
ick
nes
sIa
cob
elli
set
al.
2008
b4
7M
ult
iple
246
4556
Ad
ult
Tre
atm
ent
Pan
elII
IE
cho
Th
ick
nes
sM
alav
azo
set
al.
2008
42
Ital
y27
330
No
ne
Ech
oT
hic
kn
ess
De
Vo
set
al.
2008
53
Net
her
lan
ds
573
670
Ad
ult
Tre
atm
ent
Pan
elII
IC
TV
olu
me
Wil
len
s,et
al.
2007
54
US
A23
4435
No
ne
Ech
oT
hic
kn
ess
Iaco
bel
lis
etal
.20
053
5It
aly
3041
33N
on
eE
cho
Th
ick
nes
sIa
cob
elli
set
al.
2003
36
Ital
y72
4750
No
ne
Ech
oT
hic
kn
ess
Dif
fere
nt
asp
ects
of
on
est
ud
yw
ere
rep
ort
edse
par
atel
y38,4
9b
ut
com
bin
edh
ere.
aH
IVca
ses
excl
ud
ed.
MR
I,m
agn
etic
reso
nan
ceim
agin
g;
CT
,co
mp
ute
dto
mo
gra
ph
y;
NA
,n
ot
avai
lab
le.
33
FIG. 2. Correlation coefficient and 95% confidence intervals (CI) for each of the studies that evaluated the relationshipbetween epicardial adipose tissue (EAT) and body mass index (BMI). The mean correlation coefficient, lower limit, upperlimit, and significance level for the entire group are shown in the last line. The Cochrane Q statistic as an assessment ofheterogeneity is shown at the bottom left.
FIG. 3. Correlation coefficient and 95% confidence intervals (CI) for each of the studies that evaluated the relationshipbetween epicardial adipose tissue (EAT) and waist circumference (WC). The mean correlation coefficient, lower limit, upperlimit, and significance level for the entire group are shown in the last line. The Cochrane Q statistic as an assessment ofheterogeneity is shown at the bottom left.
34 RABKIN
Considering that there were different methodologies em-ployed for measurement of EAT, the relationship was ex-amined for the 17 studies that had echocardiographic EATmeasurements that showed a significant correlation(r = 0.474, p < 0.0000, n = 1171) between EAT and BMI. Thenine studies with EAT measured by computed tomography(CT) or magnetic resonance imaging (MRI) (the data werecombined because of the smaller number of studies usingeach alone) also showed a significant (P < 0.0001) correlation(r = 0.437) between EAT and BMI (n = 1319). There was nosignificant (z = 1.16) difference between the correlation ofEAT and BMI between the two different approaches mea-suring EAT, i.e., echocardiography or CT/MRI.
There were 20 studies that presented data with the uni-variate correlation coefficient between EAT and WC19–39
(Fig. 3). Each study showed a significant correlation, so thecombination of all studies indicated a highly significant(P < 0.00001) correlation between EAT and WC with a totalsample size of 2168 persons. Sensitivity analysis showed thatexclusion of any single study from the analysis did not sig-nificantly alter the overall findings. In addition, there werefour studies that presented the data as a significance level,and all four reported a highly significant correlation betweenEAT and WC.15–17,53 Thus, all studies reported a significantrelationship between EAT (regardless of the method of itsmeasurement) and WC. The correlation between EAT andWC (r = 0.567) was significantly (z = - 4.57, P < 0.0001) greaterthan the correlation between EAT and BMI (r = 0.469).
There were 11 studies that correlated EAT andVAT15,19,23–25,34,36,38–42 (Fig. 4). Ten studies presented thedata, and each of them found a significant correlation;thus, the combination of all studies indicated a highlysignificant (P < 0.00001) correlation between EAT andVAT. The total sample size was 570 persons. Sensitivityanalysis showed that exclusion of any single study from theanalysis did not significantly alter the overall findings. Inaddition, another study that did not present the actual cor-
relation coefficient found a significant relationship betweenEAT and VAT.15 The correlation between EAT and VAT(r = 0.0.686) was significantly (z = - 4.18, P < 0.0001) greaterthan the correlation between EAT and WC. Most studiesmeasuring VAT also measured BMI. Exclusion of two studies,with a sample size of 78 persons, that did not present data forboth EAT and BMI or VAT did not alter the relationship be-tween EAT and VAT. VAT was measured primarily by twodifferent methods MRI or CT [including dual-energy X-rayabsorptiometry (DEXA)]. The correlation of EAT with VATwas significant regardless of the method of VAT measurementand was significantly (z = - 2.59; P < 0.01) higher in studiesthat used CT compared to those that used MRI (r = 0.766 vs.0.648, respectively).
There were 11 studies that correlated SBP with EAT (Fig.5). Eight of the studies showed a significant correlation be-tween EAT and SBP. Overall there was a highly significant(P < 0.00001) correlation between EAT and SBP. The totalsample size was 1395 persons. Sensitivity analysis showedthat exclusion of any single study from the analysis did notsignificantly alter the overall findings. Although there was astatistically significant relationship between EAT and SBP,the correlation coefficient was not high (r = 0.216). The cor-relation between EAT and BMI was significantly (P < 0.001)greater than the correlation between EAT and SBP.
There were 15 studies that correlated TGs and EAT (Fig.6). Eleven of the studies showed a significant correlationbetween EAT and TGs. The total sample size was 2287 per-sons and there was a highly significant (P < 0.00001) corre-lation between EAT and TGs, although the correlationcoefficient was not high (r = 0.292). Sensitivity analysisshowed that exclusion of any single study from the analysisdid not significantly alter the overall findings.
There were 13 studies that correlated HDL and EAT (Fig.7). The correlation coefficients were always negative, indi-cating a lower HDL with greater EAT. Nine of the studiesshowed a significant correlation between EAT, and overall
FIG. 4. Correlation coefficient and 95% confidence intervals (CI) for each of the studies that evaluated the relationshipbetween epicardial adipose tissue (EAT) and visceral adipose tissue (VAT). The mean correlation coefficient, lower limit,upper limit, and significance level for the entire group are shown in the last line. The Cochrane Q statistic as an assessment ofheterogeneity is shown at the bottom left.
EAT AND METABOLIC SYNDROME 35
there was a highly significant ( p < 0.00001) correlation be-tween EAT and HDL. The total sample size was 1749 per-sons. Sensitivity analysis showed that exclusion of any singlestudy from the analysis did not significantly alter the overallfindings.
There were 12 studies that correlated FBG and EAT (Fig.8). Seven of the studies showed a significant correlation be-tween EAT and FBG. Overall there was a highly significant
(P < 0.0001) correlation between EAT and FBG with a corre-lation coefficient (r = 0.240). The total sample size was 1412persons. Sensitivity analysis showed that exclusion of anysingle study from the analysis did not significantly alter theoverall findings. Overall, there was a significant relationshipbetween EAT and FBG, but the proportion of studies that didnot find a relationship is larger than the other correlationanalysis.
FIG. 6. Correlation coefficient and 95% confidence intervals (CI) for each of the studies that evaluated the relationshipbetween epicardial adipose tissue (EAT) and triglycerides. The mean correlation coefficient, lower limit, upper limit, andsignificance level for the entire group are shown in the last line. The Cochrane Q statistic as an assessment of heterogeneity isshown at the bottom left.
FIG. 5. Correlation coefficient and 95% confidence intervals (CI) for each of the studies that evaluated the relationshipbetween epidcardial adipose tissue (EAT) and systolic blood pressure (SBP). The mean correlation coefficient, lower limit,upper limit, and significance level for the entire group are shown in the last line. The Cochrane Q statistic as an assessment ofheterogeneity is shown at the bottom left.
36 RABKIN
EAT in metabolic syndrome
Eighteen studies examined EAT in relation to the meta-bolic syndrome, with most (82%) using the National Cho-lesterol Education Program (NCEP) definition9 and the
minority using the International Diabetes or the Harmo-nized definition of metabolic syndrome8,43 (Table 1). EATwas compared in persons with or without metabolic syn-drome.33,41,44–49 Echo-measured EAT was the only mea-surement used because the units were consistent rather than
FIG. 8. Correlation coefficient and 95% confidence intervals (CI) for each of the studies that evaluated the relationshipbetween epicardial adipose tissue (EAT) and fasting blood glucose (FBG). The mean correlation coefficient, lower limit, upperlimit, and significance level for the entire group are shown in the last line. The Cochrane Q statistic as an assessment ofheterogeneity is shown at the bottom left.
FIG. 7. Correlation coefficient and 95% confidence intervals (CI) for each of the studies that evaluated the relationshipbetween epicardial adipose tissue (EAT) and high-density lipoprotein cholesterol (HDL-C). The mean correlation coefficient,lower limit, upper limit, and significance level for the entire group are shown in the last line. The Cochrane Q statistic as anassessment of heterogeneity is shown at the bottom left.
EAT AND METABOLIC SYNDROME 37
the variations in EAT measurement of thickness, area, orvolume reported by studies using CT or MRI to evaluateEAT. There were six studies that reported mean – standarddeviation (SD) and permitted pooling of the data (Fig. 9).One study that reported the median and range (without SD)is included for comparison both alone as well as in combi-nation of studies when considering the mean values only.47
All studies except one33 used the NCEP definition9; how-ever, there was no difference in the data for that study.There was more EAT in persons with the metabolic syn-drome (n = 427) because EAT was 7.5 – 0.1 mm in thicknesscompared to 4.0 – 0.1 mm in controls (n = 301). Meta-analysis comparing the SD for the difference showed ahighly significant ( p < 0.000) difference between persons withmetabolic syndrome compared to those without metabolicsyndrome. Sensitivity analysis showed that exclusion of anysingle study from the analysis did not significantly alter theoverall findings. One study used MRI to measure EAT
thickness in the anterior intraventricular groove and founda significantly greater EAT thickness in persons with meta-bolic syndrome compared to controls.49 The values forEAT thickness (9.3 – 2.5 vs. 5.8 – 1.8 mm) were similar to thestudies using echocardiographically measured EAT.49
Multivariate analysis
The relationship between EAT and metabolic syndromeafter adjusting for relevant factors in multivariate analysis isnot clear. Studies that asked a limited question in a generalpopulation group, namely whether EAT was related to thecomponents of the metabolic syndrome, usually found a sig-nificant relationship32,46,50,51 In contrast, EAT was not relatedto metabolic syndrome factors in other studies,26,29,52 e.g.,patients with glucose intolerance in one study.39 Yorgun et al.found in multiple regression analysis that mean EAT thick-ness correlated with metabolic syndrome, age and BMI but
FIG. 9. Epicardial adipose tissue (EAT) thickness for each of the studies where it was measured by echocardiography inpersons with and without the metabolic syndrome. MetS, metabolic syndrome; SD, standard deviation.
Table 2. Relationship Between Epicardial Fat and the Metabolic Syndrome in Studies
Using Multivariate Analysis and the Other Variables Included
Factors adjusted for
Relationship Age Gender WC BMI Other
Yorgun et al. 201344 Significant Yes Yes No NoKerr et al. 201352 Not significant No + No No Yes Diabetes mellitus, IL-6, coronary calcificationYerramasu et al. 201226 Not significant No + Yes No Yes Diabetes mellitus, IL-6, LDL-C, CRPSironi et al. 201223 Not significant No No No NoPierdomenico et al. 201146 Significant Yes No Yes YesWang CP et al. 200950 Significant Yes Yes No NoWang TD 200932 Significant Yes Yes Yes YesOkyay et al. 200851 Significant Yes Yes Yes No
+ Not significant in univariate analysis.WC, waist circumference; BMI, body mass index; IL-6, interleukin-6; LDL-C, low-density lipoprotein cholesterol; CRP, C-reactive protein.
38 RABKIN
not TGs or HDL.44 De Vos found a significant independentcorrelation of EAT with age and WC, but no additional sig-nificant relationship between SBP, HDL, TG, or FBG.53 If thenumber of variables was expanded to include other factors,such as the presence of diabetes mellitus, its duration as wellas other inflammatory markers such as interleukin-6 (IL-6),the strength of the association between EAT and the metabolicsyndrome decreased markedly and was not significant52
(Table 2).
Discussion
This meta-analysis demonstrates convincingly that EATsignificantly correlates with the principal clinical indicatorsof obesity, specifically BMI and WC. Importantly, there is astronger correlation between EAT and abdominal adiposity,measured by MRI, CT, or DEXA, than with the other indi-cators of obesity. EAT also correlates significantly with four(nonobesity) components of the metabolic syndrome, spe-cifically FBG, SBP, TGs, and HDL-C. These correlations,however, were not as strong as the correlation between EATand indices of obesity. Multivariate analysis supports a re-lationship between EAT and the metabolic syndrome, inde-pendent of obesity, but the relationship is not robust.
The majority of studies showed a significant relationshipbetween EAT and BMI, but the few exceptions are worthdiscussion. The reasons for those exceptions are uncertain,but one possibility is that the relationship is not valid inmorbid obesity because two of the studies that failed to finda significant association consisted of persons with morbidobesity.42,54 This proposed explanation is consistent withthe concept that there is a limitation to the space into whichEAT can expand, in contrast to the relatively minimalconstraints for increases in generalized adiposity. It hasbeen suggested that there is a sex difference in the rela-tionship of EAT to BMI, with a relationship mainly or ex-clusively in men.17 That suggestion is not substantiated bythis meta-analysis, which showed a significant relationshipbetween EAT and BMI with a sex distribution that was al-most equal (55% men). Furthermore the relationship be-tween EAT and BMI was evident even in a study thatconsisted only of women.21
Evaluation of the data across studies demonstrated a highcorrelation between EAT and VAT that was significantlystronger than the relationship between EAT and BMI or WC.This finding has several implications. The data providecompelling evidence for a linkage between the two adiposedepots (EAT and VAT). Anatomic similarities in animals, inthe absence of detailed embryologic investigation, suggesteda common origin during embryogenesis of EAT and VAT.55
This is an intriguing but as yet unproved suggestion. Data inthe present study showing the consistency of the EAT andVAT relationship between studies, as well as the high cor-relation, strongly suggests that the two fat depots share acommonality of features or characteristics.
The relationship between EAT and VAT was significantlygreater than the relationship between EAT and WC. WC canbe confounded by subcutaneous fat, which becomes an in-creasing problem with obese individuals. The higher corre-lation between EAT and VAT buttresses the argument thatmeasurement of EAT on echocardiography maybe a usefuland less expensive method to assess VAT compared to CT,MRI, or DEXA.
EAT was almost twice as thick in persons with metabolicsyndrome compared to those without metabolic syndrome.Echocardiographic measurement was used for consistency. Thedata from some MRI studies support this finding, and indeedthe 2:1 ratio was almost identical to the ratio found on MRI.19
EAT has utility in the diagnosis of metabolic syndrome, and ithas been suggested that an EAT 9.5 mm or greater in men and7.5 mm or greater in women has a high sensitivity and speci-ficity for prediction of metabolic syndrome.47
VAT has been implicated as playing a major role in pro-duction of hypertension, dyslipidemia, and glucose intoler-ance, which increase cardiovascular risk.56 The significantrelationship between EAT and factors such as blood glucosemost likely reflects the association of VAT and insulin re-sistance or glucose intolerance.35 The strong correlation be-tween EAT and VAT raises the question of whether EAT isalso responsible for the production of the other componentsof the metabolic syndrome or whether the larger amount ofadipose tissue mass in the abdomen is the dominant factor.Until interventions can specifically target only one fat depot,this question will likely be unresolved.
The metabolic syndrome is constructed from several dif-ferent elements but with slight variations in the diagnosticcriteria.8,9,43 Each of three different metabolic syndrome cri-teria was used in the studies that comprised this report.Several studies have shown that there is a significant rela-tionship between the greater the amount of EAT and thelarger the number of criteria that constitute the metabolicsyndrome.44,51
Some data suggest that the specific location of EAT, i.e., itsregional distribution over the heart, may vary in the strengthof the association with the metabolic syndrome, with astronger association evident with EAT on the right and leftatrioventricular groove and anterior right ventricle thanother areas of the heart.44 A precise explanation for thispossibility, however, is not available.
The limitations of this meta-analysis mainly relate to thenature of the studies evaluated. The kinds of studies varyfrom the baseline characteristics of morbid obesity personswho are having bariatric surgery to surveys of individualswith normal weight. The spectrum of studies, however,strengthens the ability to extrapolate the data. Second,studies published in non-English languages were excludedfrom our meta-analysis. However, excluding non-Englishstudies does not materially alter most meta-analyses.57 Third,the studies did not have a common protocol, such as bloodpressure, for collection or reporting of patient characteristics.However, this is a standard issue with meta-analysis. Fourth,the imaging techniques for assessment of EAT vary. Themajority, almost two-thirds, of the studies used echocardi-ography to measure EAT. Thus, here is considerable con-sistency. Despite the differences in EAT measurementmethodology, the results are relatively similar betweenstudies. This is due in part to the similarity in measurementbetween techniques because CT or MRI correlate well withthe echocardiography measurement, the most frequentlyused methodology in EAT assessment.49 Nevertheless, thereis concern about the accuracy and standardization of EATmeasurement both within and between the different imagingmodalities, including echocardiography, MRI, and CT.Fourth, although CT provides high spatial resolution andtrue volume coverage of the heart, there is no standard al-gorithm for data acquisition, interpretation, and EAT
EAT AND METABOLIC SYNDROME 39
quantification, and data are insufficient regarding the normaldistribution of fat volumes in the population, thus cautioningthe application of even this technology to EAT measure-ment.58 Fifth, it is important to note that a high proportion ofthe studies included in the meta-analysis were in populationswith the metabolic syndrome, obesity, or diabetes. Last, thereis a question of whether the data from this study can beextrapolated to the general population because of the num-ber of studies that included persons with the metabolicsyndrome in this study. However, studies that recruitedpersons without consideration of the presence or metabolicsyndrome or had a high proportion of persons from thegeneral population showed the same relationship betweenEAT and obesity or cardiovascular risk factors.17,59
This review focused on EAT and did not include peri-cardial fat because of the differences in the location of the fatdepots and the potential for problems with data interpreta-tion when combining two different fat depots. Nevertheless,it is important to note that in population studies such as theFramingham Heart Study, there is a higher prevalence for allcardiovascular risk factors, including diabetes, low HDL-C,hypertension, and metabolic syndrome in categories of highintrathoracic and pericardial fat deposits.60 In the Multi-Ethnic Study of Atherosclerosis (MESA) population, peri-cardial fat was significantly correlated with both BMI andWC and was also correlated with the risk of developingcoronary heart disease after further adjustment for BMI andother CVD risk factors.61
In summary, EAT correlates with the degree of obesity asdetermined by BMI; however, EAT has a significantly stron-ger association with visceral adiposity than BMI or WC. EATcorrelates significantly with each of the (other) components ofthe metabolic syndrome— SBP, TGs, HDL-C, or FBG—to asimilar extent that is considerably less than the relationship ofEAT to BMI. The very strong correlation between EAT andVAT suggests a relationship between these two adipose tissuedepots. Echocardiographic measurement of EAT can be usefulto indicate VAT. Although EAT correlates significantly witheach of the components of the metabolic syndrome (SBP, TGs,HDL-C, or FBG), the magnitude of the relationship is con-siderably and significantly less than the relationship of EAT toBMI. These data raise the possibility that the association ofEAT with CAD cannot be readily explained by standard riskfactors, and the unique features of this adipose tissue warrantsfurther investigations.
Author Disclosure Statement
The author declares no conflicts of interest.
References
1. Rabkin SW, Mathewson FA, Hsu PH. Relation of bodyweight to development of ischemic heart disease in a cohortof young North American men after a 26 year observationperiod: The Manitoba Study. Am J Cardiol 1977;39:452–458.
2. Rabkin SW. Epicardial fat: Properties, function and rela-tionship to obesity. Obes Rev 2007;8:253–261.
3. Xu Y, Cheng X, Hong K, Huang C, et al. How to interpretepicardial adipose tissue as a cause of coronary artery dis-ease: A meta-analysis. Coron Artery Dis 2012; 23:227–233.
4. Bornfeldt KE, Tabas I. Insulin resistance, hyperglycemia,and atherosclerosis. Cell Metab 2011;14:575–585.
5. Landsberg L, Aronne LJ, Beilin LJ, et al. Obesity-relatedhypertension: Pathogenesis, cardiovascular risk, and treat-ment: A position paper of The Obesity Society and theAmerican Society of Hypertension. J Clin Hypertens 2013;15:14–33.
6. Subramanian S, Chait A. Hypertriglyceridemia secondary toobesity and diabetes. Biochim Biophys Acta 2012;1821:819–825.
7. Haas MJ, Mooradian AD. Regulation of high-density lipo-protein by inflammatory cytokines: Establishing links be-tween immune dysfunction and cardiovascular disease.Diabetes Metab Res Rev 2010;26:90–99.
8. Alberti KG, Eckel RH, Grundy SM, et al. International Dia-betes Federation Task Force on Epidemiology and Preven-tion. Hational Heart and Blood Institute.American HeartAssociation.Wor L. Harmonizing the metabolic syndrome: ajoint interim statement of the International Diabetes Fed-eration Task Force on Epidemiology and Prevention; Na-tional Heart, Lung, and Blood Institute; American HeartAssociation; World Heart Federation; International. Circu-lation 2009;120:1640–1645.
9. Grundy SM, Cleeman JI, Daniels SR, et al. American HeartAssociation. National Heart and Blood Institute L. Diagnosisand management of the metabolic syndrome: An AmericanHeart Association/National Heart, Lung, and Blood In-stitute Scientific Statement. Circulation 2005;112:2735–2752.
10. Mottillo S, Filion KB, Genest J, et al. The metabolic syndromeand cardiovascular risk a systematic review and meta-analysis. J Am Coll Cardiol 2010;56:1113–1132.
11. Galassi A, Reynolds K, He J. Metabolic syndrome and risk ofcardiovascular disease: A meta-analysis. Am J Med 2006;119:812–819.
12. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis ofobservational studies in epidemiology: A proposal for re-porting. Meta-analysis Of Observational Studies in Epide-miology (MOOSE) group. JAMA 2000;283:2008–2012.
13. Moher D, Liberati A, Tetzlaff J, et al. Preferred reportingitems for systematic reviews and meta-analyses: The PRIS-MA statement. J Clin Epidemiol 2009;62:1006–1012.
14. Field AP. Meta-analysis of correlation coefficients: A MonteCarlo comparison of fixed- and random-effects methods.Psychol Methods 2001;6:161–180.
15. Bettencourt N, Toschke AM, Leite D, et al. Epicardial adi-pose tissue is an independent predictor of coronary athero-sclerotic burden. Int J Cardiol 2012;158:26–32.
16. Kim M-K, Tanaka K, Kim M-J, et al. Epicardial fat tissue:Relationship with cardiorespiratory fitness in men. Med SciSports Exerc 2010;42:463–469.
17. Lai Y-H, Yun C-H, Yang F-S, et al. Epicardial adipose tissuerelating to anthropometrics, metabolic derangements andfatty liver disease independently contributes to serum high-sensitivity C-reactive protein beyond body fat composition:A study validated with computed tomography. J Am SocEchocardiogr 2012;25:234–241.
18. Gaborit B, Jacquier A, Kober F, et al. Effects of bariatricsurgery on cardiac ectopic fat: Lesser decrease in epicardialfat compared to visceral fat loss and no change in myocar-dial triglyceride content. J Am Coll Cardiol 2012;60:1381–1389.
19. Graner M, Siren R, Nyman K, et al. Cardiac steatosis asso-ciates with visceral obesity in nondiabetic obese men. J ClinEndocrinol Metab 2013;98:1189–1197.
20. Borruel S, Fernandez-Duran E, Alpanes M, et al. Globaladiposity and thickness of intraperitoneal and mesentericadipose tissue depots are increased in women with poly-
40 RABKIN
cystic ovary syndrome (PCOS). J Clin Endocrinol Metab2013;98:1254–1263.
21. Kilicaslan B, Ozdogan O, Aydin M, et al. Increased epicardialfat thickness is associated with cardiac functional changes inhealthy women. Tohoku J Exp Med 2012;228:119–124.
22. Cavalcante JL, Tamarappoo BK, Hachamovitch R, et al. As-sociation of epicardial fat, hypertension, subclinical coronaryartery disease, and metabolic syndrome with left ventriculardiastolic dysfunction. Am J Cardiol 2012;110:1793–1798.
23. Sironi AM, Petz R, De Marchi D, et al. Impact of increasedvisceral and cardiac fat on cardiometabolic risk and disease.Diabet Med 2012;29:622–627.
24. Graner M, Seppala-Lindroos A, Rissanen A, et al. Epicardialfat, cardiac dimensions, and low-grade inflammation inyoung adult monozygotic twins discordant for obesity.American J Cardiol 2012;109:1295–1302.
25. Gaborit B, Kober F, Jacquier A, et al. Assessment of epicar-dial fat volume and myocardial triglyceride content in se-verely obese subjects: Relationship to metabolic profile,cardiac function and visceral fat. Int J Obes 2012;36:422–430.
26. Yerramasu A, Dey D, Venuraju S, et al. Increased volume ofepicardial fat is an independent risk factor for acceleratedprogression of sub-clinical coronary atherosclerosis. Athero-sclerosis 2012;220:223–230.
27. Karadag B, Ozulu B, Ozturk FY, et al. Comparison of epi-cardial adipose tissue (EAT) thickness and anthropometricmeasurements in metabolic syndrome (MS) cases above andunder the age of 65. Arch Gerontol Geriatr 2011;52:e79–e84.
28. Iacobellis G, Lonn E, Lamy A, et al. Epicardial fat thicknessand coronary artery disease correlate independently ofobesity. Int J Cardiol 2011;146:452–454.
29. Sicari R, Sironi AM, Petz R, et al. Pericardial rather thanepicardial fat is a cardiometabolic risk marker: An MRI vsecho study. J Am Soc Echocardiogr 2011;24:1156–1162.
30. Malavazos AE, Di Leo G, Secchi F, et al. Relation of echo-cardiographic epicardial fat thickness and myocardial fat.Am J Cardiol 2010;105:1831–1835.
31. Aydin H, Toprak A, Deyneli O, et al. Epicardial fat tissuethickness correlates with endothelial dysfunction and othercardiovascular risk factors in patients with metabolic syn-drome. Metab Syndr Relat Disord 2010;8:229–234.
32. Wang TD, Lee WJ, Shih FY, et al. Relations of epicardialadipose tissue measured by multidetector computed to-mography to components of the metabolic syndrome areregion-specific and independent of anthropometric indexesand intraabdominal visceral fat. J Clin Endocrinol Metab2009;94:662–669.
33. Okyay K, Balcioglu AS, Tavil Y, et al. A relationship betweenechocardiographic subepicardial adipose tissue and metabolicsyndrome. Int J Cardiovasc Imaging 2008;24:577–583.
34. Iacobellis G, Pellicelli AM, Grisorio B, et al. Relation ofepicardial fat and alanine aminotransferase in subjects withincreased visceral fat. Obesity (Silver Spring) 2008;16:179–183.
35. Iacobellis G, Leonetti F. Epicardial adipose tissue and insulinresistance in obese subjects. J Clin Endocrinol Metab 2005;90:6300–6302.
36. Iacobellis G, Ribaudo MC, Assael F, et al. Echocardiographicepicardial adipose tissue is related to anthropometric andclinical parameters of metabolic syndrome: A new indicatorof cardiovascular risk. J Clin Endocrinol Metab 2003;88:5163–5168.
37. Nakazato R, Rajani R, Cheng VY, et al. Weight change mod-ulates epicardial fat burden: A 4-year serial study withnon-contrast computed tomography. Atherosclerosis 2012;220:139–144.
38. Fu C-P, Sheu WH-H, Lee I-T, et al. Effects of weight loss onepicardial adipose tissue thickness and its relationship be-tween serum soluble CD40 ligand levels in obese men. ClinChim Act 2013;421:98–103.
39. Sironi AM, Petz R, De Marchi D, et al. Impact of increasedvisceral and cardiac fat on cardiometabolic risk and disease.Diabet Med 2012;29:622–627.
40. Gaborit B, Jacquier A, Kober F, et al. Effects of bariatricsurgery on cardiac ectopic fat: Lesser decrease in epicardialfat compared to visceral fat loss and no change in myocar-dial triglyceride content. J Am Coll Cardiol 2012;60:1381–1389.
41. Stramaglia G, Greco A, Guglielmi G, et al. Echocardio-graphy and dual-energy x-ray absorptiometry in the elderlypatients with metabolic syndrome: A comparison of twodifferent tecniques to evaluate visceral fat distribution. JNutrHealth Aging 2010;14:6–10.
42. Malavazos AE, Ermetici F, Cereda E, et al. Epicardial fatthickness: Relationship with plasma visfatin and plasmino-gen activator inhibitor-1 levels in visceral obesity. Nutr Me-tab Cardiovasc Dis 2008;18:523–530.
43. Alberti KG, Zimmet P, Shaw J. IDF Epidemiology TaskForce Consensus G. The metabolic syndrome—a newworldwide definition. Lancet 2005;366:1059–1062.
44. Yorgun H, Canpolat U, Hazirolan T, et al. Increased epi-cardial fat tissue is a marker of metabolic syndrome in adultpatients. Int J Cardiol 2013;165:308–313.
45. Sengul C, Duman D. The association of epicardial fatthickness with blunted heart rate recovery in patients withmetabolic syndrome. Tohoku J Exp Med 2011;224:257–262.
46. Pierdomenico SD, Pierdomenico AM, Neri M, et al. Epi-cardial adipose tissue and metabolic syndrome in hyper-tensive patients with normal body weight and waistcircumference. Am J Hypertens 2011;24:1145–1149.
47. Iacobellis G, Willens HJ, Barbaro G, et al. Threshold valuesof high-risk echocardiographic epicardial fat thickness.Obesity (Silver Spring) 2008;16:887–892.
48. Tok D, Cagli K, Kadife I, et al. Impaired coronary flow re-serve is associated with increased echocardiographic epi-cardial fat thickness in metabolic syndrome patients. CoronArtery Dis 2013;24:191–195.
49. Liang KW, Tsai IC, Lee WJ, et al. MRI measured epicardialadipose tissue thickness at the right AV groove differentiatesinflammatory status in obese men with metabolic syndrome.Obesity 2012;20:525–532.
50. Wang CP, Hsu HL, Hung WC, et al. Increased epicardialadipose tissue (EAT) volume in type 2 diabetes mellitus andassociation with metabolic syndrome and severity of coro-nary atherosclerosis. Clin Eendocrinol 2009;70:876–882.
51. Okyay K, Balcioglu AS, Tavil Y, et al. A relationship betweenechocardiographic subepicardial adipose tissue and metabolicsyndrome. Int J Cardiovasc Imaging 2008;24:577–583.
52. Kerr JD, Holden RM, Morton AR, et al. Associations ofepicardial fat with coronary calcification, insulin resistance,inflammation, and fibroblast growth factor-23 in stage 3–5chronic kidney disease. BMC Nephrol 2013;14:26–30.
53. De Vos AM, Prokop M, Roos CJ, et al. Peri-coronary epi-cardial adipose tissue is related to cardiovascular risk factorsand coronary artery calcification in post-menopausal wo-men. Eur Heart J 2008;29:777–783.
54. Willens HJ, Byers P, Chirinos JA, et al. Effects of weight lossafter bariatric surgery on epicardial fat measured usingechocardiography. Am J Cardiol 2007;99:1242–1245.
55. Marchington JM, Mattacks CA, Pond CM. Adipose tissue inthe mammalian heart and pericardium: Structure, foetal
EAT AND METABOLIC SYNDROME 41
development and biochemical properties. Comp BiochemPhysiol B 1989;94:225–232.
56. Despres J-P. Body fat distribution and risk of cardiovasculardisease: An update. Circulation 2012;126:1301–1313.
57. Juni P, Holenstein F, Sterne J, et al. Direction and impact oflanguage bias in meta-analyses of controlled trials: Empiricalstudy. Int J Epidemiol 2002;31:115–123.
58. Marwan M, Achenbach S. Quantification of epicardial fat bycomputed tomography: Why, when and how? J CardiovascComput Tomogr 2013;7:3–10.
59. Nakazato R, Rajani R, Cheng VY, et al. Weight changemodulates epicardial fat burden: A 4-year serial study withnon-contrast computed tomography. Atherosclerosis. 2012;220:139–144.
60. Thanassoulis G, Massaro JM, Hoffmann U, et al. Prevalence,distribution, and risk factor correlates of high pericardialand intrathoracic fat depots in the Framingham heart study.Circ Cardiovasc Imaging 2010;3:559–566.
61. Ding J, Hsu FC, Harris TB, et al. The association of peri-cardial fat with incident coronary heart disease: The Multi-Ethnic Study of Atherosclerosis (MESA). Am J Clin Nutr2009;90:499–504.
62. Mazzoccoli G, Dagostino MP, Fontana A, et al. Influence ofthe Gly1057Asp variant of the insulin receptor substrate 2(IRS2) on insulin resistance and relationship with epicardialfat thickness in the elderly. Exp Gerontol 2012;47:988–993.
63. Shemirani H, Khoshavi M. Correlation of echocardiographicepicardial fat thickness with severity of coronary arterydisease-an observational study. Anadolu Kardiyol Derg 2012;12:200–205.
64. Gorter PM, van Lindert AS, de Vos AM, et al. Quantificationof epicardial and peri-coronary fat using cardiac computedtomography; reproducibility and relation with obesity andmetabolic syndrome in patients suspected of coronary arterydisease. Atherosclerosis 2008;197:896–903.
Address correspondence to:Simon Rabkin, MD, FRCPC, FACC
University of British Columbia2775 Laurel Street, 9th Floor
Vancouver, British Columbia, V5Z 1M9Canada
E-mail: [email protected]
42 RABKIN